This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
TheraP was a randomized, parallel-group, open-label phase 2 treatment trial comparing 177Lu-PSMA617 with cabazitaxel in participants with progressive metastatic castration resistant prostate cancer. The registry reports an enrollment of 201 participants and one primary efficacy endpoint: prostate specific antigen response rate.
| Feature | TheraP |
|---|---|
| Trial acronym | TheraP |
| ClinicalTrials.gov identifier | NCT03392428 |
| Phase | Phase 2 |
| Status | Completed |
| Population | Progressive metastatic castration resistant prostate cancer |
| Conditions | Cancer of the Prostate; Metastatic Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 201 |
| Number of arms | 2 |
| Lead sponsor | Australian and New Zealand Urogenital and Prostate Cancer Trials Group |
2. Clinical Question
The clinical question was whether treatment with 177Lu-PSMA617 should be compared with cabazitaxel in participants with progressive metastatic castration resistant prostate cancer, using prostate specific antigen response rate as the primary endpoint.
Population
Participants with progressive metastatic castration resistant prostate cancer.
Intervention
177Lu-PSMA617.
Comparator
Cabazitaxel.
Primary question
How does the proportion of participants achieving a PSA reduction of at least 50% from baseline compare between the two randomized groups?
3. Trial Design
177Lu-PSMA617
- 177Lu-PSMA617
- Randomized treatment assignment
Cabazitaxel
- Cabazitaxel
- Randomized treatment assignment
Randomization is the central design feature for the comparative question. By assigning participants prospectively to treatment groups, the design creates the basis for comparing outcomes between the groups rather than simply describing outcomes among patients who independently received different therapies.
The absence of masking is also statistically relevant. The treatment assignment is known rather than concealed from participants and investigators. For a PSA-based endpoint, the primary outcome is defined using a quantitative laboratory measure, which can reduce some forms of subjective outcome assessment even though an open-label design can still influence other aspects of trial conduct.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Prostate Specific Antigen response rate (PSA RR) | PSA RR defined as the proportion of participants in each group with a PSA reduction of ≥ 50% from baseline. | Through study completion, on average 4 years |
The primary endpoint is a binary response endpoint. Each participant is classified according to whether the prespecified PSA response criterion is achieved: a reduction of at least 50% from baseline. The resulting treatment-group quantity is a proportion.
The response rate summarizes the fraction of participants meeting the registry's ≥50% PSA reduction criterion. The treatment comparison concerns these proportions, rather than a comparison of mean PSA values.
5. Planned Analysis
The ClinicalTrials.gov record identifies prostate specific antigen response rate as the primary endpoint but does not report posted statistical analyses or numerical efficacy results.
Primary endpoint
Response is defined as a PSA reduction of ≥ 50% from baseline.
Time frame: through study completion, on average 4 years.
For an endpoint defined as the proportion of participants meeting a binary response criterion, the natural statistical analysis compares the response proportions between the randomized treatment groups. Depending on the prespecified statistical analysis plan, this can be expressed using a difference in proportions, a relative measure such as a risk ratio, or an odds ratio, together with a confidence interval and an appropriate hypothesis test.
The key statistical distinction is between estimating each response rate and testing whether the groups differ. A response rate by itself is descriptive. A formal comparative analysis incorporates uncertainty and the randomized treatment assignment.
| Statistical quantity | What it answers |
|---|---|
| Response proportion | What fraction of participants met the ≥50% PSA reduction criterion? |
| Difference in proportions | How many percentage points separate the two treatment groups? |
| Risk ratio | How many times as frequent was response in one group relative to the other? |
| Odds ratio | How do the odds of response compare between the two groups? |
| Confidence interval | How precisely is the treatment contrast estimated? |
| Hypothesis test | How compatible are the observed data with a specified null comparison? |
6. Statistical Methodology
Binary endpoint analysis
The PSA response endpoint converts the continuous underlying PSA measurement into a binary outcome using a clinically defined threshold: at least a 50% reduction from baseline. This makes the primary endpoint a proportion rather than a continuous change score.
For each randomized group, the analysis begins with two counts: the number of participants meeting the response definition and the number of participants included in the response assessment. The response rate is their ratio. The treatment comparison then evaluates the difference between those rates.
Here, x denotes the number of responders and n denotes the number of participants evaluated in each treatment group.
Difference in response rates
A difference in proportions provides an absolute treatment contrast. If one group has response probability p1 and the other has response probability p2, the absolute difference is:
An absolute difference is expressed in proportion units or percentage points and directly describes the separation between the two response rates.
This measure is often clinically intuitive because it describes an absolute change in the probability of meeting the response definition. It does not, however, describe the magnitude of PSA reduction among responders or the duration of the response.
Risk ratio
A risk ratio compares the response probability in one group with the response probability in the other. A value of 1 corresponds to equal response probabilities; values above or below 1 indicate relative differences in the probability of response, depending on which group is placed in the numerator.
Unlike an absolute difference, the risk ratio is a relative measure. It should therefore be interpreted alongside the underlying response rates.
Odds ratio
An odds ratio compares the odds of response rather than the probabilities directly. It is mathematically useful for binary outcomes but can be less intuitive than a risk difference or risk ratio when response is common.
The odds ratio should not be described as if it were a risk ratio. When the outcome is not rare, the numerical values can differ substantially.
Confidence intervals for response comparisons
A confidence interval supplies information about statistical precision. For a treatment difference, the interval describes the uncertainty around the estimated difference under the chosen statistical framework. For a ratio measure, the interval describes uncertainty on the ratio scale.
The confidence interval should be considered together with the point estimate. A narrow interval indicates greater precision than a wide interval, while the width depends on factors such as the number of participants and the number of observed responses.
Hypothesis testing
A formal test for a binary endpoint typically starts with a null hypothesis specifying no treatment difference on the chosen effect scale. The resulting p-value addresses compatibility of the observed data with that null hypothesis under the statistical model and testing procedure.
A p-value is not a measure of treatment effect size. Two studies can have the same estimated treatment difference but different p-values if their sample sizes or variability differ. Conversely, a small p-value does not by itself establish that an observed difference is large or clinically important.
7. Statistical Methods Explained
Why is PSA response rate a binary endpoint?
The registry defines response using a threshold: a PSA reduction of at least 50% from baseline. Participants therefore fall into two categories for the primary endpoint—meeting the response criterion or not meeting it. This simplifies a continuous biomarker trajectory into an endpoint that can be compared as a proportion.
What does a response rate actually measure?
The response rate measures the proportion of participants who achieve the prespecified ≥50% PSA reduction. It does not measure the average PSA reduction across all participants, the amount of reduction beyond 50%, or how long an individual response persists.
Why compare proportions rather than means?
The registered endpoint is explicitly defined as a proportion of participants. A comparison of means would answer a different question about the underlying PSA measurements. The statistical analysis should therefore preserve the binary structure of the registered endpoint.
What is the difference between an absolute and relative treatment effect?
An absolute effect such as a difference in response rates describes the number of percentage points separating the groups. A relative effect such as a risk ratio describes the response probability in one group relative to the other. Both can be informative, but they answer different questions.
Why does the confidence interval matter?
The point estimate is only one estimate from the observed trial. A confidence interval communicates the precision of that estimate under the specified statistical framework. A wider interval indicates greater uncertainty than a narrower interval; it does not mean that every value inside the interval is equally plausible.
Why does the p-value not measure treatment benefit?
A p-value evaluates the evidence against a specified null hypothesis. It does not quantify how large the treatment effect is, nor does it describe the probability that a treatment is effective. Effect size and statistical evidence should therefore be reported separately.
8. Analysis Population and Randomization
The registry identifies the study as randomized but does not provide detailed analysis-population definitions in the record summarized here. This distinction matters because a binary response analysis can produce different denominators if participants are excluded from the response assessment.
Randomized comparison
Randomization provides the framework for comparing the two assigned treatment groups.
Response denominator
The response rate depends on which participants are included in the response assessment denominator.
Missing assessments
The registry does not report a missing-data or imputation method for the primary endpoint.
Interpretive consequence
The analysis population and handling of unavailable PSA assessments can affect the estimated response proportion.
For a randomized trial, the intention-to-treat principle is often important because retaining participants according to their randomized assignment preserves the original comparison. For a response endpoint, however, the exact estimand and handling of participants without an evaluable response assessment must be defined prospectively. The ClinicalTrials.gov record does not provide those additional details.
9. Time Frame and Endpoint Interpretation
The registry gives the primary endpoint time frame as through study completion, on average 4 years. This identifies the period over which the primary endpoint is assessed in the registry description.
Study start
The registered study start date was January 29, 2018.
Primary completion
The registered primary completion date was December 31, 2019.
Primary endpoint time frame
PSA response rate was defined through study completion, on average 4 years.
A time frame is important because a response rate is always tied to a defined assessment period. A response observed early in treatment and a response observed later are not interchangeable unless the analysis specifies how timing is handled. The registry's wording establishes the endpoint's overall time frame but does not provide a detailed schedule of PSA assessments.
10. What the Primary Endpoint Does — and Does Not — Mean
The primary endpoint asks whether a participant achieved a PSA reduction of ≥ 50% from baseline. The resulting response rate is the proportion of participants satisfying that threshold.
A PSA response is not the same as overall survival, progression-free survival, duration of response, or a measure of how much PSA changed in participants who did not cross the 50% threshold.
Threshold endpoints intentionally simplify continuous information. Two participants who both meet the response definition may have different PSA reductions, while two participants just above and just below the threshold can be classified differently despite relatively similar measurements.
When results are eventually reported, the response proportion in each randomized group should be shown directly. Relative measures can then be interpreted against those absolute rates rather than in isolation.
11. Statistical Interpretation of a Future Treatment Contrast
Because the registry contains no posted statistical analyses, there is no numerical treatment-effect estimate to interpret. The appropriate framework can nevertheless be stated clearly.
| Measure | Interpretive question |
|---|---|
| 177Lu-PSMA617 response rate | What proportion achieved a PSA reduction of ≥50% from baseline? |
| Cabazitaxel response rate | What proportion achieved a PSA reduction of ≥50% from baseline? |
| Absolute difference | How many percentage points separate the two response rates? |
| Relative measure | How does the probability or odds of response compare between groups? |
| 95% confidence interval | How precise is the estimated treatment contrast? |
| P-value | How compatible are the data with the specified null hypothesis? |
The most informative presentation would place the two response rates beside the chosen treatment contrast and its confidence interval. This allows readers to see both the absolute magnitude of response and the uncertainty around the comparison.
12. Missing Data and Assessment Issues
The registry description specifies the response definition but does not report a formal missing-data or imputation method. This is an important methodological consideration for a binary endpoint because a participant without an evaluable PSA assessment cannot automatically be treated as equivalent to a participant who definitively failed to meet the response criterion.
Possible analytical choices can change the estimand. For example, an analysis restricted to participants with evaluable assessments answers a different question from an analysis that treats unavailable assessments according to a prespecified conservative rule. The appropriate choice depends on the protocol and statistical analysis plan.
13. Multiplicity and Interim Analysis
The registry information identifies one registered primary endpoint: prostate specific antigen response rate. It does not report an interim-analysis procedure, alpha-spending method, multiplicity strategy, or endpoint hierarchy beyond that primary endpoint description.
Primary endpoint
PSA response rate is the registered primary endpoint.
Interim analysis
No interim statistical analysis procedure is reported in the registry information.
Multiplicity
No multiplicity-adjustment procedure is reported in the registry information.
Interpretive principle
Formal type I error interpretation depends on the prespecified testing framework when multiple hypotheses or repeated looks at the data are involved.
A single primary endpoint simplifies the conceptual testing framework, but it does not by itself establish how secondary endpoints, repeated assessments, subgroup analyses, or interim looks were handled. Those features require their own prespecified statistical specifications.
14. Non-Inferiority, Bayesian Methods, and Stratification
The registry information does not identify a non-inferiority margin, Bayesian statistical method, or stratification factors for the primary analysis. These concepts therefore do not form part of the documented statistical framework available here.
| Design topic | Registry information |
|---|---|
| Non-inferiority margin | Not reported |
| Bayesian methods | Not reported |
| Stratification factors | Not reported |
| Interim analysis | Not reported |
| Multiplicity procedure | Not reported |
| Missing-data / imputation method | Not reported |
The distinction is important because these methods answer different design questions. Non-inferiority trials require a prespecified margin; Bayesian analyses require a prior and posterior framework; stratified analyses require defined strata; and multiplicity procedures address the inflation of type I error across multiple statistical tests.
15. Safety
The registry information identifies the two treatment arms but does not provide serious adverse event counts or other numerical safety results in the record summarized here.
| Safety information | Registry record |
|---|---|
| 177Lu-PSMA617 serious adverse events by arm | Not reported |
| Cabazitaxel serious adverse events by arm | Not reported |
| Numerical safety comparison | Not reported |
Safety should be analyzed separately from the PSA response endpoint. A treatment-group difference in response rate does not quantify adverse-event risk, and a safety comparison does not substitute for an efficacy analysis. In a complete clinical-trial assessment, efficacy and safety are complementary dimensions of the randomized comparison.
16. Trial History
Study begins
The registered study start date was January 29, 2018.
Primary completion
The registered primary completion date was December 31, 2019.
Study status
The ClinicalTrials.gov record lists the study status as completed.
17. Limitations
- No posted numerical efficacy analysis: the registry identifies the primary endpoint but does not provide a treatment-effect estimate, confidence interval, or p-value.
- Limited endpoint detail: the registry defines PSA response as a ≥50% reduction from baseline but does not provide a detailed assessment schedule.
- Analysis population: detailed definitions of the population used for the primary response analysis are not reported in the registry information.
- Missing data: the registry does not report how unavailable or unevaluable PSA assessments were handled statistically.
- Statistical testing framework: the registry does not report the specific hypothesis test, effect measure, confidence-interval method, or multiplicity procedure used for the primary endpoint.
- Safety results: numerical serious adverse event results by treatment arm are not reported in the registry information.
- Generalizability: the registered population is specifically characterized by progressive metastatic castration resistant prostate cancer, so the statistical comparison is tied to that clinical setting.
These limitations do not prevent understanding the trial's basic statistical structure. They do limit how far the quantitative analysis can proceed using the ClinicalTrials.gov record alone. The most important available information is the randomized two-arm design, the primary PSA response definition, and the specified time frame.
18. Why This Trial Matters Statistically
TheraP provides a useful teaching example of how a randomized clinical trial can use a binary biomarker-defined response endpoint rather than a time-to-event endpoint. The statistical questions are consequently different from trials centered on overall survival or progression-free survival.
| Concept | How it appears in TheraP |
|---|---|
| Randomization | Participants were randomly allocated to two treatment groups. |
| Parallel-group design | The trial used a two-arm parallel design. |
| Open-label design | Masking was registered as none. |
| Binary endpoint | PSA response is defined by whether a participant achieves a ≥50% reduction from baseline. |
| Response rate | The primary endpoint is the proportion of participants meeting the response criterion. |
| Absolute treatment effect | A difference in response proportions expresses the separation in percentage points. |
| Relative treatment effect | A risk ratio or odds ratio can express the treatment contrast on a relative scale. |
| Confidence interval | Quantifies precision around the estimated treatment contrast. |
| Hypothesis testing | A formal test evaluates the observed treatment difference against a specified null hypothesis. |
| Missing-data considerations | The response denominator and handling of unevaluable assessments can affect the analysis. |
19. Statistical Concepts in This Trial
The TheraP design illustrates several foundational principles of clinical-trial statistics. A binary endpoint changes the analysis problem from estimating a time-to-event distribution to estimating and comparing response probabilities. The resulting interpretation is most transparent when the response rates, absolute treatment difference, relative treatment measure, and confidence interval are considered together.
Binary outcomes
A participant either meets the registered ≥50% PSA response criterion or does not.
Absolute effects
A difference in response rates describes the separation between groups in percentage-point terms.
Relative effects
Risk ratios and odds ratios describe the treatment comparison on relative scales.
Statistical precision
Confidence intervals show how precisely a treatment contrast has been estimated.
20. Record Summary
TheraP is a completed randomized phase 2 parallel-group treatment trial with 201 participants and two interventions: 177Lu-PSMA617 and cabazitaxel. The registered primary endpoint is prostate specific antigen response rate, defined as the proportion of participants in each group with a PSA reduction of ≥ 50% from baseline, assessed through study completion, on average 4 years.
Statistically, the primary endpoint is a binary response outcome. The fundamental analysis compares response proportions between the randomized groups. An absolute difference describes the percentage-point separation, while a relative measure such as a risk ratio or odds ratio describes the treatment contrast on a relative scale. Confidence intervals quantify precision, while p-values address compatibility with a specified null hypothesis rather than measuring effect size.
The ClinicalTrials.gov record does not post formal statistical analyses or numerical primary-endpoint results. It also does not report a detailed missing-data method, stratification scheme, interim-analysis procedure, multiplicity strategy, non-inferiority margin, Bayesian method, or numerical serious adverse event results by arm.